Starbucks Capstone Challenge

Project Overview

In this project, I will try to find how Starbucks customers use the app, and how well is the current offers system. I will also see who should the app target in promotions. The data sets used in this project contains simulated data that mimics customer behavior on the Starbucks rewards mobile app. From it, we can understand the costumers' behavior and it might help us make better decisions.

The Problem

The problem we have here is that we don't want to give any customer our offers. We want to give only those who we think will be able to complete the offer. Giving an offer to someone we know he/she probably will not be able to complete it is a waste of time and resources that can be given to someone who we know will complete it. I will approach this problem by first cleaning up the data, then doing some exploratory analysis and see who are my most valuable customers after that I will create a model to help us predicting feature customers and which type of offer should we give them.

Analysis

Part I: Business understanding

My objective here is to find patterns and show when and where to give specific offer to a specific customer. Main users of this kind of applications are Starbucks employees and analysts. The plan in this project to have questions and answer them with data visualization. Tha data is provided by Starbucks contains simulated data that mimics customer behavior.

Part II: Data Exploration

In this project we were given 3 files. Before I start analyzing we have to explore and see what is the data we have. We need to check if it is clean or not, if each column have the right type that the data tell, for example if the data in column called price is saved as string, we need to convert it to number to help us in the analysis if we want to find the sum for example, having it as string will not return the total of that column. Similar thing goes to dates saved as strings.

# import libraries
import pandas as pd
import numpy as np
import math
import json
import plotly
import plotly.graph_objs as go
import matplotlib.pyplot as plt
import seaborn as sns
plotly.offline.init_notebook_mode()

%matplotlib inline
from sklearn.preprocessing import MultiLabelBinarizer
import warnings
warnings.filterwarnings('ignore')

# read in the json files
portfolio = pd.read_json('data/portfolio.json', orient='records', lines=True)
profile = pd.read_json('data/profile.json', orient='records', lines=True)
transcript = pd.read_json('data/transcript.json', orient='records', lines=True)

Part III: Data understanding

The data we have is provided by Starbucks. Here is a quick breakthrough of how the data looks like:

• three .json files.
  • portfolio.json - containing offer ids and meta data about each offer (duration, type, etc.)
  • profile.json - demographic data for each customer
  • transcript.json - records for transactions, offers received, offers viewed, and offers completed Here is the schema and explanation of each variable in the files:

portfolio.json

• id (string) - offer id
• offer_type (string) - type of offer ie BOGO, discount, informational
• difficulty (int) - minimum required spend to complete an offer
• reward (int) - reward given for completing an offer
• duration (int) - time for offer to be open, in days
• channels (list of strings)

portfolio.sample(5)

	reward	channels	difficulty	duration	offer_type	id
3	5	[web, email, mobile]	5	7	bogo	9b98b8c7a33c4b65b9aebfe6a799e6d9
1	10	[web, email, mobile, social]	10	5	bogo	4d5c57ea9a6940dd891ad53e9dbe8da0
8	5	[web, email, mobile, social]	5	5	bogo	f19421c1d4aa40978ebb69ca19b0e20d
5	3	[web, email, mobile, social]	7	7	discount	2298d6c36e964ae4a3e7e9706d1fb8c2
4	5	[web, email]	20	10	discount	0b1e1539f2cc45b7b9fa7c272da2e1d7

portfolio.shape

(10, 6)

profile.json

• age (int) - age of the customer
• became_member_on (int) - date when customer created an app account
• gender (str) - gender of the customer (note some entries contain 'O' for other rather than M or F)
• id (str) - customer id
• income (float) - customer's income

profile.sample(5)

	gender	age	id	became_member_on	income
3680	F	65	44eae499188b4c5c849741ac698aed4b	20161216	97000.0
11260	None	118	ea796e04a85942e3a77985ed1f9a1940	20170420	NaN
13033	M	29	17a83b17fa7b4293989b830cfa30546a	20171121	38000.0
14310	M	20	bd52aa1e72284ae69f476817c9989794	20180123	39000.0
15943	M	65	6b1c9e0cc8a54186a4129131a1726738	20180105	100000.0

profile.shape

(17000, 5)

transcript.json

• event (str) - record description (ie transaction, offer received, offer viewed, etc.)
• person (str) - customer id
• time (int) - time in hours since start of test. The data begins at time t=0
• value - (dict of strings) - either an offer id or transaction amount depending on the record

transcript.sample(5)

	person	event	value	time
187613	ce0d0f8ee78448568754d6f9d8c90b1f	transaction	{'amount': 20.41}	456
165717	b2aaf8311ae9422981ded134c2bbbc34	offer viewed	{'offer id': '4d5c57ea9a6940dd891ad53e9dbe8da0'}	408
290871	feca5113ed4f47268bcde4c176c6dd9f	transaction	{'amount': 1.03}	648
210912	75d9de6a8e66420ba4215e53003e0f42	offer received	{'offer id': '4d5c57ea9a6940dd891ad53e9dbe8da0'}	504
137747	f277d871eb28447bb86e4b73a9bc4379	transaction	{'amount': 30.4}	366

transcript.shape

(306534, 4)

Part IV: Data preparation / Wrangling

In this part, I will go though the data and do some data wrangling and fix some issues with coulmns such as Value in the transcript table, and Channels in the portfolio and others.

First list check data types

portfolio

	reward	channels	difficulty	duration	offer_type	id
0	10	[email, mobile, social]	10	7	bogo	ae264e3637204a6fb9bb56bc8210ddfd
1	10	[web, email, mobile, social]	10	5	bogo	4d5c57ea9a6940dd891ad53e9dbe8da0
2	0	[web, email, mobile]	0	4	informational	3f207df678b143eea3cee63160fa8bed
3	5	[web, email, mobile]	5	7	bogo	9b98b8c7a33c4b65b9aebfe6a799e6d9
4	5	[web, email]	20	10	discount	0b1e1539f2cc45b7b9fa7c272da2e1d7
5	3	[web, email, mobile, social]	7	7	discount	2298d6c36e964ae4a3e7e9706d1fb8c2
6	2	[web, email, mobile, social]	10	10	discount	fafdcd668e3743c1bb461111dcafc2a4
7	0	[email, mobile, social]	0	3	informational	5a8bc65990b245e5a138643cd4eb9837
8	5	[web, email, mobile, social]	5	5	bogo	f19421c1d4aa40978ebb69ca19b0e20d
9	2	[web, email, mobile]	10	7	discount	2906b810c7d4411798c6938adc9daaa5

portfolio.dtypes

reward         int64
channels      object
difficulty     int64
duration       int64
offer_type    object
id            object
dtype: object

One issue with the portfolio dataframe is having a list of items in the channels column, to fix this one I will do something similar to one-hot-encoding and make three new columns for each type of channels and put 1 when it applies on that promotion and 0 if it doesn't apply.

mlb = MultiLabelBinarizer()
res = pd.DataFrame(mlb.fit_transform(portfolio['channels']),
                   columns=mlb.classes_,
                   index=portfolio['channels'].index)
portfolio = portfolio.drop('channels', axis=1)
portfolio = pd.concat([portfolio, res], axis=1, sort=False)
portfolio

	reward	difficulty	duration	offer_type	id	email	mobile	social	web
0	10	10	7	bogo	ae264e3637204a6fb9bb56bc8210ddfd	1	1	1	0
1	10	10	5	bogo	4d5c57ea9a6940dd891ad53e9dbe8da0	1	1	1	1
2	0	0	4	informational	3f207df678b143eea3cee63160fa8bed	1	1	0	1
3	5	5	7	bogo	9b98b8c7a33c4b65b9aebfe6a799e6d9	1	1	0	1
4	5	20	10	discount	0b1e1539f2cc45b7b9fa7c272da2e1d7	1	0	0	1
5	3	7	7	discount	2298d6c36e964ae4a3e7e9706d1fb8c2	1	1	1	1
6	2	10	10	discount	fafdcd668e3743c1bb461111dcafc2a4	1	1	1	1
7	0	0	3	informational	5a8bc65990b245e5a138643cd4eb9837	1	1	1	0
8	5	5	5	bogo	f19421c1d4aa40978ebb69ca19b0e20d	1	1	1	1
9	2	10	7	discount	2906b810c7d4411798c6938adc9daaa5	1	1	0	1

For portfolio dataframe, we can see that we don't have NaN values.

profile.head()

	gender	age	id	became_member_on	income
0	None	118	68be06ca386d4c31939f3a4f0e3dd783	20170212	NaN
1	F	55	0610b486422d4921ae7d2bf64640c50b	20170715	112000.0
2	None	118	38fe809add3b4fcf9315a9694bb96ff5	20180712	NaN
3	F	75	78afa995795e4d85b5d9ceeca43f5fef	20170509	100000.0
4	None	118	a03223e636434f42ac4c3df47e8bac43	20170804	NaN

profile.dtypes

gender               object
age                   int64
id                   object
became_member_on      int64
income              float64
dtype: object

First we can notice that we have some NaN values in the gender and income, lets find the sum of in all columns.

profile.isna().sum()

gender              2175
age                    0
id                     0
became_member_on       0
income              2175
dtype: int64

As we can see we have some NaN values in gender and income columns, for gender I will fill NaNs with N/A, and for income I will fill NaNs with the mean.

profile['gender'].fillna('NA', inplace=True)
profile['income'].fillna((profile['income'].mean()), inplace=True)

While working through the data I notices that there is age number 118 which seems impossible to happen and to make sure I also noticed that for all profiles that have age equal to 118, they also don't have gender listed so it might be written wrong or it is the default value. For those values I will keep 118 as it is.

profile.isna().sum()

gender              0
age                 0
id                  0
became_member_on    0
income              0
dtype: int64

profile.dtypes

gender               object
age                   int64
id                   object
became_member_on      int64
income              float64
dtype: object

Lastly lets check transcript dataframe

transcript.head()

	person	event	value	time
0	78afa995795e4d85b5d9ceeca43f5fef	offer received	{'offer id': '9b98b8c7a33c4b65b9aebfe6a799e6d9'}	0
1	a03223e636434f42ac4c3df47e8bac43	offer received	{'offer id': '0b1e1539f2cc45b7b9fa7c272da2e1d7'}	0
2	e2127556f4f64592b11af22de27a7932	offer received	{'offer id': '2906b810c7d4411798c6938adc9daaa5'}	0
3	8ec6ce2a7e7949b1bf142def7d0e0586	offer received	{'offer id': 'fafdcd668e3743c1bb461111dcafc2a4'}	0
4	68617ca6246f4fbc85e91a2a49552598	offer received	{'offer id': '4d5c57ea9a6940dd891ad53e9dbe8da0'}	0

transcript.dtypes

person    object
event     object
value     object
time       int64
dtype: object

Checking for NaN values we can see that we have none, we have to clean the value column since it hold dictonary of offer id, amount or reward.

transcript.isna().sum()

person    0
event     0
value     0
time      0
dtype: int64

# find different keys in value column
keys = []
for idx, row in transcript.iterrows():
    for k in row['value']:
        if k in keys:
            continue
        else:
            keys.append(k)

keys

['offer id', 'amount', 'offer_id', 'reward']

As we can see above there are offer_id and offer id, after fixing this we need to concatinate those columns togther since they are the same.

# Iterate over transcript table, check value column and update it, put each key in seperated column.
transcript['offer_id'] = ''
transcript['amount'] = 0
transcript['reward'] = 0

for idx, row in transcript.iterrows():
    for k in row['value']:
        if k == 'offer_id' or k == 'offer id':
            transcript.at[idx, 'offer_id']= row['value'][k]
        if k == 'amount':
            transcript.at[idx, 'amount']= row['value'][k]
        if k == 'reward':
            transcript.at[idx, 'reward']= row['value'][k]

# example
transcript.loc[306506]

person                       b895c57e8cd047a8872ce02aa54759d6
event                                         offer completed
value       {'offer_id': 'fafdcd668e3743c1bb461111dcafc2a4...
time                                                      714
offer_id                     fafdcd668e3743c1bb461111dcafc2a4
amount                                                      0
reward                                                      2
Name: 306506, dtype: object

Now no need to keep the value column and we can drop it.

transcript = transcript.drop('value', axis=1)
transcript.head()

	person	event	time	offer_id	amount	reward
0	78afa995795e4d85b5d9ceeca43f5fef	offer received	0	9b98b8c7a33c4b65b9aebfe6a799e6d9	0	0
1	a03223e636434f42ac4c3df47e8bac43	offer received	0	0b1e1539f2cc45b7b9fa7c272da2e1d7	0	0
2	e2127556f4f64592b11af22de27a7932	offer received	0	2906b810c7d4411798c6938adc9daaa5	0	0
3	8ec6ce2a7e7949b1bf142def7d0e0586	offer received	0	fafdcd668e3743c1bb461111dcafc2a4	0	0
4	68617ca6246f4fbc85e91a2a49552598	offer received	0	4d5c57ea9a6940dd891ad53e9dbe8da0	0	0

Now, the data looks ready for some analysis and machine learning.

Part V: Analysis / Modeling

Divided to two parts, Analysis and Modeling, I will start with Analysis where I will analyze the data and try to find insights.

1. Analysis:

A. Univariate Exploration:

1- How many varibales do I have in each dataframe? What are the types?
2- What are the most common values for each column in each dataframe?
3- What is the average income for Starbucks customers?
4- What is the average age for Starbucks customers?
5- What is the most common promotion?
6- What is the least common promotion?
7- Who are the most loyal customer (most transcripts)?
8- What are the most events we have in our transcripts?

B. Multivariate Exploration:

1- What is the most common promotion for children, teens, adult and elderly customors?
2- From profiles, which get more income, males or females?
3- What is the gender distribution in the transcript dataframe?
4- Who takes long time to achieve each promotion goal and from which gender, age, income?
5- How many customers we get each month (became_member_on)?
6- What is the average length between two transcript for the same customer? 
7- Which type of promotions each gender likes (offer_type)?
8- From each offer received by customer, how many they completed?

A. Univariate Exploration:

A1: How many varibales do I have in each dataframe? What are the types?

We already looked into the shape and check of NaNs and viewed the type of each column. Lets see for numerical and categorical data, how many each appeared, what is the mean and min/max .. etc.

profile.head(5)

	gender	age	id	became_member_on	income
0	NA	118	68be06ca386d4c31939f3a4f0e3dd783	20170212	65404.991568
1	F	55	0610b486422d4921ae7d2bf64640c50b	20170715	112000.000000
2	NA	118	38fe809add3b4fcf9315a9694bb96ff5	20180712	65404.991568
3	F	75	78afa995795e4d85b5d9ceeca43f5fef	20170509	100000.000000
4	NA	118	a03223e636434f42ac4c3df47e8bac43	20170804	65404.991568

profile.describe()

	age	became_member_on	income
count	17000.000000	1.700000e+04	17000.000000
mean	62.531412	2.016703e+07	65404.991568
std	26.738580	1.167750e+04	20169.288288
min	18.000000	2.013073e+07	30000.000000
25%	45.000000	2.016053e+07	51000.000000
50%	58.000000	2.017080e+07	65404.991568
75%	73.000000	2.017123e+07	76000.000000
max	118.000000	2.018073e+07	120000.000000

plt.figure(figsize=(13, 4))
sns.boxplot(profile['age'])
plt.title('Age Boxplot')
plt.xlabel('Age')
plt.xticks(rotation = 90)
plt.show();

Form the about box plot, we can see that most of ages in our profile dataframe falls in-between 40 and 80. We already notice one outlier which is 118. Our median is around 58 years old.

What about income?

plt.figure(figsize=(16, 4))
sns.boxplot(profile['income'])
plt.title('Income Boxplot')
plt.xlabel('income')
plt.xticks(rotation = 90)
plt.show();

Our boxplot shows that the median is around 65k and most of incomes falls between 50k and 78k.

A2: What are the most common values for each column in each dataframe?

For Age, we have 85 different ages, so to make the graph look better I think we should divide ages to groups:

• Child: 0 - 12.
• Teen: 12 - 18.
• Young Adult: 18 - 21.
• Adult: 21 - 64.
• Elderly: 65+.

profile['age'].value_counts().shape[0]

85

# reference: https://twitter.com/justmarkham/status/1146040449678925824
profile['age_groups'] = pd.cut(profile.age, bins=[0, 12, 18, 21, 64, 200], labels=['child', 'teen', 'young adult', 'adult', 'elderly'])

# I am ignoring the outlier age 118
plt.figure(figsize=(16, 4))
top10_ages = profile['age'].value_counts()[1:].head(10).reset_index()
plt.bar(top10_ages['index'], top10_ages['age'])
plt.title('Common Ages in Profiles')
plt.ylabel('Number of Profiles')
plt.xlabel('Age')
plt.xticks(top10_ages['index'], rotation = 0)
plt.show();

sns.countplot(x='age_groups', data=profile)
plt.title('Number of Profiles In Each Age Group')
plt.ylabel('Number of Profiles')
plt.xlabel('Age Group')
plt.xticks(rotation = 45)
plt.show();

sns.countplot(profile['gender'])
plt.title('Genders in Profiles')
plt.ylabel('Total')
plt.xlabel('Gender')
plt.xticks(rotation = 0)
plt.show();

We can see that we have alot profile in the adult age group, ages between 21 and 64, lets see more about this and look inside the adult age group and see how it is divided.

adults = profile[profile['age_groups'] == 'adult']

plt.figure(figsize=(16, 6))
sns.countplot(adults['age'])
plt.title('Number of Profiles In Each Age')
plt.ylabel('Number of Profiles')
plt.xlabel('Age')
plt.xticks(rotation = 45)
plt.show();

For become_member_on column, because we have alot of dates, I will group it by month.

profile['became_member_on_month'] = profile['became_member_on'].astype(str).str[:6]

plt.figure(figsize=(16, 6))
sns.countplot(profile['became_member_on_month'])
plt.title('Number of Profiles In Each Month')
plt.ylabel('Number of Profiles')
plt.xlabel('Year-Month')
plt.xticks(rotation = 90)
plt.show();

We can see that we did better job between August 2017 and January 2018. We were able to get more than 800 new profiles monthly

For transcript dataframe, lets see the common values we have.

sns.countplot(transcript['event'])
plt.title('Number of events In Transcripts')
plt.ylabel('Number of Transcripts')
plt.xlabel('Transcript type')
plt.xticks(rotation = 0)
plt.show();

We can see that most of the transcripts are transactions. Around 75% of the offer received were viewed. And nearly 50% of the viewed offers were completed.

A3: What is the average income for Starbucks customers?

profile['income'].describe()['mean']

65404.99156829799

A4: What is the average age for Starbucks customers?

profile['age'].describe()['mean']

62.53141176470588

A5: What are the most common promotion? and what are their types?

Here I will be looking for only completed promotions.

completed_off_count = transcript[transcript['event'] == 'offer completed']

sns.countplot(y=completed_off_count['offer_id'])
plt.title('Number of Completed Promotion for each Offer')
plt.ylabel('Promotion ID')
plt.xticks(rotation = 45)
plt.show();

print(f'Offer ID: {completed_off_count["offer_id"].value_counts().index[0]}')
print(f'Number of Completion: {completed_off_count["offer_id"].value_counts().values[0]}')

Offer ID: fafdcd668e3743c1bb461111dcafc2a4
Number of Completion: 5317

Now lets look at the most common types of offers, to find that I need to get the offer type from the portfolio dataframe.

def get_offer_type(offer_id):
    try:
        offer_type = portfolio[portfolio['id'] == offer_id]['offer_type'].values[0]
        return offer_type
    except:
        offer_type = 'NA'
        return offer_type

transcript['offer_type'] = transcript.apply(lambda x: get_offer_type(x['offer_id']), axis=1)

sns.countplot(transcript[transcript['offer_type'] != 'NA']['offer_type'])
plt.title('Number of Completed Promotion for Type Offer')
plt.ylabel('Number of transactions')
plt.xlabel('Offer Type')
plt.xticks(rotation = 45)
plt.show();

As we saw in the above graph, it is pretty close between BOGO offers and discount offers.

A6: What is the least common promotion?

print(f'Offer ID: {completed_off_count["offer_id"].value_counts().index[-1]}')
print(f'Number of Completion: {completed_off_count["offer_id"].value_counts().values[-1]}')

Offer ID: 4d5c57ea9a6940dd891ad53e9dbe8da0
Number of Completion: 3331

A7: Who are the most loyal customer (most transcripts)?

For this one, I will check offer completed and transactions event types.

loyal_customer_count = transcript[(transcript['event'] == 'offer completed') | (transcript['event'] == 'transaction')].groupby(['person', 'event'])['amount'].sum().reset_index()
loyal_customer_count = loyal_customer_count.sort_values('amount', ascending=False).head(10)
count = 1

print(' ')
print(' *************** [ LEADERBOARD ] **************')
print(' ')
for idx, row in loyal_customer_count.iterrows():
    print(f'.------------------- [ #{count} ] ------------------.')
    print(f'| Profile ID: {row["person"]} |')
    print(f'| Number of Completed Offers:       {completed_off_count[(completed_off_count["person"] == row["person"]) & (completed_off_count["event"] == "offer completed")].shape[0]}          |')
    print(f'| Amount:                         ${row["amount"]}        |')
    print(f"'----------------------------------------------'")
    count += 1

 
 *************** [ LEADERBOARD ] **************
 
.------------------- [ #1 ] ------------------.
| Profile ID: 3c8d541112a74af99e88abbd0692f00e |
| Number of Completed Offers:       5          |
| Amount:                         $1606        |
'----------------------------------------------'
.------------------- [ #2 ] ------------------.
| Profile ID: f1d65ae63f174b8f80fa063adcaa63b7 |
| Number of Completed Offers:       6          |
| Amount:                         $1360        |
'----------------------------------------------'
.------------------- [ #3 ] ------------------.
| Profile ID: ae6f43089b674728a50b8727252d3305 |
| Number of Completed Offers:       3          |
| Amount:                         $1320        |
'----------------------------------------------'
.------------------- [ #4 ] ------------------.
| Profile ID: 626df8678e2a4953b9098246418c9cfa |
| Number of Completed Offers:       4          |
| Amount:                         $1314        |
'----------------------------------------------'
.------------------- [ #5 ] ------------------.
| Profile ID: 73afdeca19e349b98f09e928644610f8 |
| Number of Completed Offers:       5          |
| Amount:                         $1314        |
'----------------------------------------------'
.------------------- [ #6 ] ------------------.
| Profile ID: 52959f19113e4241a8cb3bef486c6412 |
| Number of Completed Offers:       5          |
| Amount:                         $1285        |
'----------------------------------------------'
.------------------- [ #7 ] ------------------.
| Profile ID: ad1f0a409ae642bc9a43f31f56c130fc |
| Number of Completed Offers:       3          |
| Amount:                         $1256        |
'----------------------------------------------'
.------------------- [ #8 ] ------------------.
| Profile ID: d240308de0ee4cf8bb6072816268582b |
| Number of Completed Offers:       5          |
| Amount:                         $1244        |
'----------------------------------------------'
.------------------- [ #9 ] ------------------.
| Profile ID: 946fc0d3ecc4492aa4cc06cf6b1492c3 |
| Number of Completed Offers:       4          |
| Amount:                         $1224        |
'----------------------------------------------'
.------------------- [ #10 ] ------------------.
| Profile ID: 6406abad8e2c4b8584e4f68003de148d |
| Number of Completed Offers:       3          |
| Amount:                         $1206        |
'----------------------------------------------'

A8: What are the most events we have in our transcripts?

sns.countplot(transcript['event'])
plt.title('Number of events In Transcripts')
plt.ylabel('Number of Transcripts')
plt.xlabel('Transcript type')
plt.xticks(rotation = 0)
plt.show();

Transaction have the most amount of rows in the transcript dataframe with around 140k, almost half of our dataframe total.

B. Multivariate Exploration:

B1: What is the most common promotion for children, teens, young adult, adult and elderly customors?

To find out, we need to get customer's age in the transcript dataframe. I created a function to get that from the profile dataframe (It takes time to run). I will ignore 'child' age group since we have no rows of it.

def get_customer_age_group(profile_id):
    age_group = profile[profile['id'] == profile_id]['age_groups'].values[0]
    return age_group

transcript['age_group'] = transcript.apply(lambda x: get_customer_age_group(x['person']), axis=1)

plt.figure(figsize=(14, 6))
sns.countplot(x="age_group", hue="offer_type", data=transcript)
plt.title('Most Popular Offers to Each Age Group')
plt.ylabel('Total')
plt.xlabel('Age Group')
plt.xticks(rotation = 0)
plt.legend(title='Offer Type')
plt.show();

B2: From profiles, which get more income, males or females?

Here I ignored those who didn't tell their gender.

plt.figure(figsize=(14, 6))
sns.violinplot(x=profile[profile['gender'] != 'NA']['gender'], y=profile['income'])
plt.title('Income vs Gender')
plt.ylabel('Income')
plt.xlabel('Gender')
plt.xticks(rotation = 0)
plt.show();

The graph above shows that income median (the white dot) for females (around 70k) is higher than males (around 60k) we can also see that for females the income spreads from 40k to 100k. For males most of them around 40k to 70k which close to median.

B3: What is the gender distribution in the transcript dataframe?

We need to add gender also here, I will make function to do so (It takes time to run).

def get_customer_gender(profile_id):
    gender = profile[profile['id'] == profile_id]['gender'].values[0]
    return gender

transcript['gender'] = transcript.apply(lambda x: get_customer_gender(x['person']), axis=1)

plt.figure(figsize=(14, 6))
sns.countplot(x=transcript[transcript["gender"] != 'NA']['gender'], hue="offer_type", data=transcript)
plt.title('Most Popular Offers to Each Gender')
plt.ylabel('Total')
plt.xlabel('Gender')
plt.xticks(rotation = 0)
plt.legend(title='Offer Type')
plt.show();

plt.figure(figsize=(14, 6))
sns.countplot(x=transcript[transcript["gender"] != 'NA']['gender'], hue="event", data=transcript)
plt.title('Most Popular Offer Event to Each Gender')
plt.ylabel('Total')
plt.xlabel('Gender')
plt.xticks(rotation = 0)
plt.legend(title='Offer Event')
plt.show();

print(f'Number of males records {transcript[transcript["gender"] == "M"].shape[0]}, and number of female records is {transcript[transcript["gender"] == "F"].shape[0]}.')

Number of males records 155690, and number of female records is 113101.

From the two graphs above we can see that males received offers more than females. Both genders seems to reflect on those offers similarly. Around half of 80% of offers received were viewed by both genders, but it seems that females would complete those offers more than males. The numbers are:

total_trans_g_o = transcript[transcript["gender"] != 'NA'].groupby(['gender','offer_type']).count()
total_trans_g_e = transcript[transcript["gender"] != 'NA'].groupby(['gender','event']).count()

total_trans_g_o, total_trans_g_e

(                      person  event   time  offer_id  amount  reward  \
 gender offer_type                                                      
 F      NA              49382  49382  49382     49382   49382   49382   
        bogo            27619  27619  27619     27619   27619   27619   
        discount        26652  26652  26652     26652   26652   26652   
        informational    9448   9448   9448      9448    9448    9448   
 M      NA              72794  72794  72794     72794   72794   72794   
        bogo            35301  35301  35301     35301   35301   35301   
        discount        34739  34739  34739     34739   34739   34739   
        informational   12856  12856  12856     12856   12856   12856   
 O      NA               1781   1781   1781      1781    1781    1781   
        bogo              914    914    914       914     914     914   
        discount          920    920    920       920     920     920   
        informational     356    356    356       356     356     356   
 
                       age_group  
 gender offer_type                
 F      NA                 49382  
        bogo               27619  
        discount           26652  
        informational       9448  
 M      NA                 72794  
        bogo               35301  
        discount           34739  
        informational      12856  
 O      NA                  1781  
        bogo                 914  
        discount             920  
        informational        356  ,
                         person   time  offer_id  amount  reward  offer_type  \
 gender event                                                                  
 F      offer completed   15477  15477     15477   15477   15477       15477   
        offer received    27456  27456     27456   27456   27456       27456   
        offer viewed      20786  20786     20786   20786   20786       20786   
        transaction       49382  49382     49382   49382   49382       49382   
 M      offer completed   16466  16466     16466   16466   16466       16466   
        offer received    38129  38129     38129   38129   38129       38129   
        offer viewed      28301  28301     28301   28301   28301       28301   
        transaction       72794  72794     72794   72794   72794       72794   
 O      offer completed     501    501       501     501     501         501   
        offer received      916    916       916     916     916         916   
        offer viewed        773    773       773     773     773         773   
        transaction        1781   1781      1781    1781    1781        1781   
 
                         age_group  
 gender event                       
 F      offer completed      15477  
        offer received       27456  
        offer viewed         20786  
        transaction          49382  
 M      offer completed      16466  
        offer received       38129  
        offer viewed         28301  
        transaction          72794  
 O      offer completed        501  
        offer received         916  
        offer viewed           773  
        transaction           1781  )

total_trans_go_o_t = total_trans_g_o.loc[('F')]['event'].sum()
total_trans_go_o_tt = total_trans_g_o.loc[('M')]['event'].sum()
total_trans_go_o_t_offers_f = total_trans_g_o.loc[('F')].loc[['bogo', 'discount', 'informational']]['event'].sum()
total_trans_go_o_t_offers_m = total_trans_g_o.loc[('M')].loc[['bogo', 'discount', 'informational']]['event'].sum()

print('For Females:')
print(f'Total transcipts is: {total_trans_go_o_t}.')
print(f"Number of bogo offers: {total_trans_g_o.loc[('F', 'bogo')].values[0]}, {round((total_trans_g_o.loc[('F', 'bogo')].values[0]/total_trans_go_o_t_offers_f)*100,2)}% of total.")
print(f"Number of discount offers: {total_trans_g_o.loc[('F', 'discount')].values[0]}, {round((total_trans_g_o.loc[('F', 'discount')].values[0]/total_trans_go_o_t_offers_f)*100,2)}% of total.")
print(f"Number of informational offers: {total_trans_g_o.loc[('F', 'informational')].values[0]}, {round((total_trans_g_o.loc[('F', 'informational')].values[0]/total_trans_go_o_t_offers_f)*100,2)}% of total.")
 
print(f"Number of offer completed: {total_trans_g_e.loc[('F', 'offer completed')].values[0]}, {round((total_trans_g_e.loc[('F', 'offer completed')].values[0]/total_trans_g_e.loc[('F', 'offer received')].values[0])*100,2)}% of total offers received.")
print(f"Number of offer received: {total_trans_g_e.loc[('F', 'offer received')].values[0]}, {round((total_trans_g_e.loc[('F', 'offer received')].values[0]/total_trans_go_o_t_offers_f)*100,2)}% of total.")
print(f"Number of offer viewed: {total_trans_g_e.loc[('F', 'offer viewed')].values[0]}, {round((total_trans_g_e.loc[('F', 'offer viewed')].values[0]/total_trans_go_o_t_offers_f)*100,2)}% of total.")
print(f"Number of transaction: {total_trans_g_e.loc[('F', 'transaction')].values[0]}, {round((total_trans_g_e.loc[('F', 'transaction')].values[0]/total_trans_go_o_t)*100,2)}% of total.")

print('\nFor Males:')
print(f'Total transcipts is: {total_trans_go_o_tt}.')
print(f"Number of bogo offers: {total_trans_g_o.loc[('M', 'bogo')].values[0]}, {round((total_trans_g_o.loc[('M', 'bogo')].values[0]/total_trans_go_o_t_offers_m)*100,2)}% of total.")
print(f"Number of discount offers: {total_trans_g_o.loc[('M', 'discount')].values[0]}, {round((total_trans_g_o.loc[('M', 'discount')].values[0]/total_trans_go_o_t_offers_m)*100,2)}% of total.")
print(f"Number of informational offers: {total_trans_g_o.loc[('M', 'informational')].values[0]}, {round((total_trans_g_o.loc[('M', 'informational')].values[0]/total_trans_go_o_t_offers_m)*100,2)}% of total.")
      
print(f"Number of offer completed: {total_trans_g_e.loc[('M', 'offer completed')].values[0]}, {round((total_trans_g_e.loc[('M', 'offer completed')].values[0]/total_trans_g_e.loc[('M', 'offer received')].values[0])*100,2)}% of total offers received.")
print(f"Number of offer received: {total_trans_g_e.loc[('M', 'offer received')].values[0]}, {round((total_trans_g_e.loc[('M', 'offer received')].values[0]/total_trans_go_o_t_offers_m)*100,2)}% of total.")
print(f"Number of offer viewed: {total_trans_g_e.loc[('M', 'offer viewed')].values[0]}, {round((total_trans_g_e.loc[('M', 'offer viewed')].values[0]/total_trans_go_o_t_offers_m)*100,2)}% of total.")
print(f"Number of transaction: {total_trans_g_e.loc[('M', 'transaction')].values[0]}, {round((total_trans_g_e.loc[('M', 'transaction')].values[0]/total_trans_go_o_t)*100,2)}% of total.")

For Females:
Total transcipts is: 113101.
Number of bogo offers: 27619, 43.34% of total.
Number of discount offers: 26652, 41.83% of total.
Number of informational offers: 9448, 14.83% of total.
Number of offer completed: 15477, 56.37% of total offers received.
Number of offer received: 27456, 43.09% of total.
Number of offer viewed: 20786, 32.62% of total.
Number of transaction: 49382, 43.66% of total.

For Males:
Total transcipts is: 155690.
Number of bogo offers: 35301, 42.58% of total.
Number of discount offers: 34739, 41.91% of total.
Number of informational offers: 12856, 15.51% of total.
Number of offer completed: 16466, 43.18% of total offers received.
Number of offer received: 38129, 46.0% of total.
Number of offer viewed: 28301, 34.14% of total.
Number of transaction: 72794, 64.36% of total.

The numbers above shows that males receive offers more than females by 9% and their transaction is 19% more too, which tells that they both more than females. Regarding offers, Males and Females received the same amount of BOGO and discount offers.

B4: Who takes long time to achieve each promotion goal and from which gender, age, income?

They are pretty close if it not similar. Both males and females take about 15 days to complete and offer.

tran_avg_len_g_f = transcript.groupby(['gender', 'offer_id'])['time'].mean().reset_index()
tran_avg_len_g_m = transcript.groupby(['gender', 'offer_id'])['time'].mean().reset_index()
tran_avg_len_g_f = tran_avg_len_g_f[tran_avg_len_g_f['offer_id'] == '']
tran_avg_len_g_m = tran_avg_len_g_m[tran_avg_len_g_m['offer_id'] == '']
print(tran_avg_len_g_f[tran_avg_len_g_f['gender'] == 'F']['time'].values[0], tran_avg_len_g_f[tran_avg_len_g_f['gender'] == 'F']['time'].values[0] / 24)
print(tran_avg_len_g_f[tran_avg_len_g_f['gender'] == 'M']['time'].values[0], tran_avg_len_g_f[tran_avg_len_g_f['gender'] == 'M']['time'].values[0] / 24)

380.8600299704346 15.869167915434774
381.72731269060637 15.905304695441933

B5: What is the average length between two transcript for the same customer?

The mean time it tekes a customer to complete an offer is around 16 days (390 hours)

tran_avg_len = transcript.groupby(['person', 'offer_id'])['time'].mean().reset_index()
tran_avg_len = tran_avg_len[tran_avg_len['offer_id'] == '']
tran_avg_len['time'].mean(), tran_avg_len['time'].mean() / 24

(390.0493165677239, 16.252054856988497)

B6: Which type of promotions each gender likes (offer_type)?

We can see that both genders like bogo and discount offers and they have the same reaction to informational offers, they both seem to be not intersted to it.

plt.figure(figsize=(14, 6))
sns.countplot(x=transcript[transcript["gender"] != 'NA']['gender'], hue="offer_type", data=transcript)
plt.title('Most Popular Offers to Each Gender')
plt.ylabel('Total')
plt.xlabel('Gender')
plt.xticks(rotation = 0)
plt.legend(title='Offer Type')
plt.show();

B7: From each offer received by customer, how many they completed?

plt.figure(figsize=(14, 6))
sns.countplot(x=transcript[transcript["gender"] != 'NA']['gender'], hue="event", data=transcript)
plt.title('Most Popular Offer Event to Each Gender')
plt.ylabel('Total')
plt.xlabel('Gender')
plt.xticks(rotation = 0)
plt.legend(title='Offer Event')
plt.show();

print('For Females:')
print(f"Number of offer completed: {total_trans_g_e.loc[('F', 'offer completed')].values[0]}, {round((total_trans_g_e.loc[('F', 'offer completed')].values[0]/total_trans_g_e.loc[('F', 'offer received')].values[0])*100,2)}% of total offers received.")
print(f"Number of offer received: {total_trans_g_e.loc[('F', 'offer received')].values[0]}, {round((total_trans_g_e.loc[('F', 'offer received')].values[0]/total_trans_go_o_t_offers_f)*100,2)}% of total.")
print(f"Number of offer viewed: {total_trans_g_e.loc[('F', 'offer viewed')].values[0]}, {round((total_trans_g_e.loc[('F', 'offer viewed')].values[0]/total_trans_go_o_t_offers_f)*100,2)}% of total.")
print(f"Number of transaction: {total_trans_g_e.loc[('F', 'transaction')].values[0]}, {round((total_trans_g_e.loc[('F', 'transaction')].values[0]/total_trans_go_o_t)*100,2)}% of total.")

print('\nFor Males:')
print(f"Number of offer completed: {total_trans_g_e.loc[('M', 'offer completed')].values[0]}, {round((total_trans_g_e.loc[('M', 'offer completed')].values[0]/total_trans_g_e.loc[('M', 'offer received')].values[0])*100,2)}% of total offers received.")
print(f"Number of offer received: {total_trans_g_e.loc[('M', 'offer received')].values[0]}, {round((total_trans_g_e.loc[('M', 'offer received')].values[0]/total_trans_go_o_t_offers_m)*100,2)}% of total.")
print(f"Number of offer viewed: {total_trans_g_e.loc[('M', 'offer viewed')].values[0]}, {round((total_trans_g_e.loc[('M', 'offer viewed')].values[0]/total_trans_go_o_t_offers_m)*100,2)}% of total.")
print(f"Number of transaction: {total_trans_g_e.loc[('M', 'transaction')].values[0]}, {round((total_trans_g_e.loc[('M', 'transaction')].values[0]/total_trans_go_o_t)*100,2)}% of total.")

For Females:
Number of offer completed: 15477, 56.37% of total offers received.
Number of offer received: 27456, 43.09% of total.
Number of offer viewed: 20786, 32.62% of total.
Number of transaction: 49382, 43.66% of total.

For Males:
Number of offer completed: 16466, 43.18% of total offers received.
Number of offer received: 38129, 46.0% of total.
Number of offer viewed: 28301, 34.14% of total.
Number of transaction: 72794, 64.36% of total.

Females completed 56% of the offers they received, it is 13% more than males, but males made more transactions than females, 64% to 43%.

Note:

Now since I did alot of changes to the transcript dataframe and it takes time to do the changes evry run, i'll convert the dataframe to csv file and just run it whenever I come back to work on the project. I will add one last column for the datafame which is income that might help us in the model (It takes time to run).

def get_customer_income(profile_id):
    income = profile[profile['id'] == profile_id]['income'].values[0]
    return income

transcript['income'] = transcript.apply(lambda x: get_customer_income(x['person']), axis=1)

transcript.to_csv('transcript_updated.csv')

transcript_up = pd.read_csv('transcript_updated.csv').iloc[:, 1:]

transcript_up.isna().sum()

person             0
event              0
time               0
offer_id      138953
amount             0
reward             0
offer_type    138953
age_group          0
gender         33772
income             0
dtype: int64

Before working on the model I will take care of the NaN values in gender, offer_type and offer_id. offer_id and offer_type are NaN because theose records are transactions, so they are not offer received, viewed or completed, because of that, we have the same number of NaNs in offer_type since they are not offers. I will replace NaNs in both columns with NA. Similarly in gender, some didn't tell their gender and we saw that before in profile dataframe, and we replaced it with NA.

fill_na = ['offer_id', 'offer_type', 'gender']
for i in fill_na:
    transcript_up[i] = transcript_up[i].fillna('NA')
transcript_up.isna().sum()

person        0
event         0
time          0
offer_id      0
amount        0
reward        0
offer_type    0
age_group     0
gender        0
income        0
dtype: int64

2. Modeling:

In this part, I will try to make a model that can identify which kind of offers we should give a customer. First lets take quick look at our final dataframe before modeling.

transcript_up.head(5)

	person	event	time	offer_id	amount	reward	offer_type	age_group	gender	income
0	78afa995795e4d85b5d9ceeca43f5fef	offer received	0	9b98b8c7a33c4b65b9aebfe6a799e6d9	0	0	bogo	elderly	F	100000.000000
1	a03223e636434f42ac4c3df47e8bac43	offer received	0	0b1e1539f2cc45b7b9fa7c272da2e1d7	0	0	discount	elderly	NA	65404.991568
2	e2127556f4f64592b11af22de27a7932	offer received	0	2906b810c7d4411798c6938adc9daaa5	0	0	discount	elderly	M	70000.000000
3	8ec6ce2a7e7949b1bf142def7d0e0586	offer received	0	fafdcd668e3743c1bb461111dcafc2a4	0	0	discount	elderly	NA	65404.991568
4	68617ca6246f4fbc85e91a2a49552598	offer received	0	4d5c57ea9a6940dd891ad53e9dbe8da0	0	0	bogo	elderly	NA	65404.991568

transcript_up.dtypes

person         object
event          object
time            int64
offer_id       object
amount          int64
reward          int64
offer_type     object
age_group      object
gender         object
income        float64
dtype: object

Because my model will guess the offer_type, I will only get those transcripts with offer id's.

transcript_up = transcript_up[transcript_up['offer_id'] != 'NA']

Now, we should split our dataframe to features and target. Our features here are:

• Event. (Will be replaced from categorical to numerical)
• Time.
• Offer_id. (Will be replaced from categorical to numerical)
• Amount.
• Reward.
• Age_group. (Will be replaced from categorical to numerical)
• Gender. (Will be replaced from categorical to numerical)
• Income.

And my target is offer_type. For my target, I will replace texts with numbers. Where BOGO = 1, discount = 2, informational = 3.

# reference: https://www.datacamp.com/community/tutorials/categorical-data
labels_event = transcript_up['event'].astype('category').cat.categories.tolist()
replace_map_comp_event = {'event' : {k: v for k,v in zip(labels_event,list(range(1,len(labels_event)+1)))}}
print(replace_map_comp_event)

labels_offer_id = transcript_up['offer_id'].astype('category').cat.categories.tolist()
replace_map_comp_offer_id = {'offer_id' : {k: v for k,v in zip(labels_offer_id,list(range(1,len(labels_offer_id)+1)))}}
print(replace_map_comp_offer_id)

labels_age_group = transcript_up['age_group'].astype('category').cat.categories.tolist()
replace_map_comp_age_group = {'age_group' : {k: v for k,v in zip(labels_age_group,list(range(1,len(labels_age_group)+1)))}}
print(replace_map_comp_age_group)

labels_gender = transcript_up['gender'].astype('category').cat.categories.tolist()
replace_map_comp_gender = {'gender' : {k: v for k,v in zip(labels_gender,list(range(1,len(labels_gender)+1)))}}
print(replace_map_comp_gender)

{'event': {'offer completed': 1, 'offer received': 2, 'offer viewed': 3}}
{'offer_id': {'0b1e1539f2cc45b7b9fa7c272da2e1d7': 1, '2298d6c36e964ae4a3e7e9706d1fb8c2': 2, '2906b810c7d4411798c6938adc9daaa5': 3, '3f207df678b143eea3cee63160fa8bed': 4, '4d5c57ea9a6940dd891ad53e9dbe8da0': 5, '5a8bc65990b245e5a138643cd4eb9837': 6, '9b98b8c7a33c4b65b9aebfe6a799e6d9': 7, 'ae264e3637204a6fb9bb56bc8210ddfd': 8, 'f19421c1d4aa40978ebb69ca19b0e20d': 9, 'fafdcd668e3743c1bb461111dcafc2a4': 10}}
{'age_group': {'adult': 1, 'elderly': 2, 'teen': 3, 'young adult': 4}}
{'gender': {'F': 1, 'M': 2, 'NA': 3, 'O': 4}}

# reference: https://www.datacamp.com/community/tutorials/categorical-data
labels_offer_type = transcript_up['offer_type'].astype('category').cat.categories.tolist()
replace_map_comp_offer_type = {'offer_type' : {k: v for k,v in zip(labels_offer_type,list(range(1,len(labels_offer_type)+1)))}}
print(replace_map_comp_offer_type)

{'offer_type': {'bogo': 1, 'discount': 2, 'informational': 3}}

# replace categorical with numerical
transcript_up.replace(replace_map_comp_event, inplace=True)
transcript_up.replace(replace_map_comp_offer_id, inplace=True)
transcript_up.replace(replace_map_comp_age_group, inplace=True)
transcript_up.replace(replace_map_comp_gender, inplace=True)

transcript_up.replace(replace_map_comp_offer_type, inplace=True)

transcript_up.head()

	person	event	time	offer_id	amount	reward	offer_type	age_group	gender	income
0	78afa995795e4d85b5d9ceeca43f5fef	2	0	7	0	0	1	2	1	100000.000000
1	a03223e636434f42ac4c3df47e8bac43	2	0	1	0	0	2	2	3	65404.991568
2	e2127556f4f64592b11af22de27a7932	2	0	3	0	0	2	2	2	70000.000000
3	8ec6ce2a7e7949b1bf142def7d0e0586	2	0	10	0	0	2	2	3	65404.991568
4	68617ca6246f4fbc85e91a2a49552598	2	0	5	0	0	1	2	3	65404.991568

# Split the data into features and target label
target = transcript_up['offer_type']
features = transcript_up.drop(['person', 'offer_type'], axis = 1)

features.head()

	event	time	offer_id	amount	reward	age_group	gender	income
0	2	0	7	0	0	2	1	100000.000000
1	2	0	1	0	0	2	3	65404.991568
2	2	0	3	0	0	2	2	70000.000000
3	2	0	10	0	0	2	3	65404.991568
4	2	0	5	0	0	2	3	65404.991568

target.head()

0    1
1    2
2    2
3    2
4    1
Name: offer_type, dtype: int64

We should normalize the numerical values ( Time, amount, reward, income ) because we will use them as features.

from sklearn.preprocessing import MinMaxScaler

scaler = MinMaxScaler()
to_normalize = ['time', 'amount', 'reward', 'income']

features[to_normalize] = scaler.fit_transform(features[to_normalize])
features.head()

	event	time	offer_id	amount	reward	age_group	gender	income
0	2	0.0	7	0.0	0.0	2	1	0.777778
1	2	0.0	1	0.0	0.0	2	3	0.393389
2	2	0.0	3	0.0	0.0	2	2	0.444444
3	2	0.0	10	0.0	0.0	2	3	0.393389
4	2	0.0	5	0.0	0.0	2	3	0.393389

Creating training and testing sets

from sklearn.model_selection import train_test_split, GridSearchCV

X_train, X_test, y_train, y_test = train_test_split(features, target, random_state=0)

print('Training Features Shape:', X_train.shape)
print('Training Labels Shape:', y_train.shape)
print('Testing Features Shape:', X_test.shape)
print('Testing Labels Shape:', y_test.shape)

Training Features Shape: (125685, 8)
Training Labels Shape: (125685,)
Testing Features Shape: (41896, 8)
Testing Labels Shape: (41896,)

Metrics:

Since we have a simple classification problem, I will use accuracy to evaluate my models. We want to see how well our model by seeing the number of correct predictions vs total number of predicitons.

I'll try different models to pick the best out of them.

1. LogisticRegression

# reference: https://towardsdatascience.com/random-forest-in-python-24d0893d51c0
def pred_score(model):
    pred = model.predict(X_test)
    # Calculate the absolute errors
    errors = abs(pred - y_test)
    # Calculate mean absolute percentage error
    mape = 100 * (errors / y_test)
    accuracy = 100 - np.mean(mape)
    return round(accuracy, 2)

from sklearn.linear_model import LogisticRegression

logreg = LogisticRegression()

logreg.fit(X_train, y_train)
print(f'Accuracy of Logistic regression classifier on training set: {round(logreg.score(X_train, y_train)*100,2)}%.')
print(f'Prediction Accuracy: {pred_score(logreg)}%')

Accuracy of Logistic regression classifier on training set: 64.32%.
Prediction Accuracy: 79.73%

2. K-Nearest Neighbors

from sklearn.neighbors import KNeighborsClassifier

knn = KNeighborsClassifier()

knn.fit(X_train, y_train)
print(f'Accuracy of K-NN classifier on training set: {round(knn.score(X_train, y_train)*100,2)}%.')
print(f'Prediction Accuracy: {pred_score(knn)}%')

Accuracy of K-NN classifier on training set: 100.0%.
Prediction Accuracy: 100.0%

3. Decision Tree

from sklearn.tree import DecisionTreeClassifier

dt = DecisionTreeClassifier()

dt.fit(X_train, y_train)
print(f'Accuracy of Decision Tree classifier on training set: {round(dt.score(X_train, y_train)*100,2)}%.')
print(f'Prediction Accuracy: {pred_score(dt)}%')

Accuracy of Decision Tree classifier on training set: 100.0%.
Prediction Accuracy: 100.0%

4. Support Vector Machine

from sklearn.svm import SVC

svm = SVC()

svm.fit(X_train, y_train)
print(f'Accuracy of SVM classifier on training set: {round(svm.score(X_train, y_train)*100,2)}%.')
print(f'Prediction Accuracy: {pred_score(svm)}%')

Accuracy of SVM classifier on training set: 91.46%.
Prediction Accuracy: 94.31%

5. Naive Bayes

from sklearn.naive_bayes import GaussianNB 

gnb = GaussianNB() 
gnb.fit(X_train, y_train) 
print(f'Accuracy of Naive Bayes classifier on training set: {round(gnb.score(X_train, y_train)*100,2)}%.')
print(f'Prediction Accuracy: {pred_score(gnb)}%')

Accuracy of Naive Bayes classifier on training set: 48.5%.
Prediction Accuracy: 25.06%

6. Random Forest

from sklearn.ensemble import RandomForestRegressor

rf = RandomForestRegressor(n_estimators = 100, random_state = 42)

rf.fit(X_train, y_train)
print(f'Accuracy of Random Forest classifier on training set: {round(rf.score(X_train, y_train)*100,2)}%.')
print(f'Prediction Accuracy: {pred_score(rf)}%')

Accuracy of Random Forest classifier on training set: 100.0%.
Prediction Accuracy: 100.0%

Models Results

# reference: https://stackoverflow.com/a/52768022
models = [logreg, knn, dt, svm, gnb, rf]
model_names = [type(n).__name__ for n in models]
tr_accuracy = [x.score(X_train, y_train)*100 for x in models]
pred_accuracy = [pred_score(y) for y in models]

results = [tr_accuracy, pred_accuracy]
results_df = pd.DataFrame(results, columns = model_names, index=['Training Accuracy', 'Predicting Accureacy']) 
results_df

	LogisticRegression	KNeighborsClassifier	DecisionTreeClassifier	SVC	GaussianNB	RandomForestRegressor
Training Accuracy	64.324303	99.997613	100.0	91.464375	48.501412	100.0
Predicting Accureacy	79.730000	100.000000	100.0	94.310000	25.060000	100.0

The above table, shows the accuracy score related with using different models of supervised learning. As presented on the table, we had 100% accuracy in both training and testing sets for 3 models ( out of 6). To avoid over fitting as much as possible, I will choose the model that gave me the middle accuracy score on the testing set,which is the Logistic Regression 79.73%. On this model , I got 79.73% on testing set. I know that this is a resonabley good score, but the other scores are higher than that (except GaussianNB). Logistic Regression is better used here since we have few binomial outcomes ( BOGO = 1, discount = 2, informational = 3 ). It is good here because we have good amount of data to work with.

Model Improvements

The result from my Logistic Regression is good enough and we got fair number and avoided overfitting. But lets try to improve it little bit.

# Tuning to get better accuracy
dual = [True, False]
max_iter = [100, 120, 140, 160]
C = [1.0,1.5]
param_grid = dict(dual = dual, max_iter = max_iter, C = C)

lr = LogisticRegression(penalty='l2')
grid = GridSearchCV(estimator = lr, param_grid = param_grid, cv = 3, n_jobs = -1)
grid_result = grid.fit(X_train, y_train)

print(f'Best Score: {grid_result.best_score_}')
print(f'Best params: {grid_result.best_params_}')

Best Score: 0.6468791025182002
Best params: {'C': 1.5, 'dual': False, 'max_iter': 120}

We improve slightly, 0.36% more. I still think it is good as it is and we don't need to try to get better results.

Althgough I believe on the saying " There is always a room for Improvement", But I think that the KNeighborsClassifier model is giving me a really good score. Trying to improve such model will surely cause us to get into the fault of Overfitting. So, I will not suggest any improvement on this model since I believe that we don't need to try to get better results.

Conclusion

In this project, I tried to analyze and make model to predict the best offer to give a Starbucks customer. First I explored the data and see what I have to change before start the analysis. Then I did some exploratory analysis on the data after cleaning. From that analysis I found out that most favorite type of offers are Buy One Get One (BOGO) offers and Discount offers. I digged deep to see who and what type of customers we have and noticed that Females tend to complete offers more than males with 56% completion of the offers they received. Where Males completed only 43.18% from the offers they received. But our current data shows that we gave males more offers since they have more transactions than females with total number of 72794 transactions, where females only had 49382 transactions. In conclusion, the company should give more offers to Females than Males since they have more completed offers. And they should focus more on BOGO and Discount offers since they are the one that tend to make customers buy more.

Improvements

I think I got to a point where we have good results and we understand the data we have very well. But to make our results even better, I would try to improve my data collection and fix issues I have with NaN values. I will also try to get even more data like location and when the transaction were completed, which branch and what time of the day. All these data can help us know when and where to give our offers. Also having more data is always good think to help us improve our model results.